skip to content

Keynes Fund


Summary of Project Plan



With the current pandemic governments are trying to persuade citizens to take certain actions based on the views of experts, from individuals adopting health precautions, to businesses owners’ keeping their businesses afloat. Major inefficiencies and coordination failures could ensue if these advices are not followed and it may be essential to prevent fears from becoming self-fulfilling: e.g., if everyone believes the stock market will crash, it will. Governments often choose which information to collect and release. How can this be best done?

We propose an online experiment to study how a representative sample of US citizens react to the opinions of experts and test theories on how to best disseminate this information. We aim to answer three questions.

  1. Which people are most responsive to experts? This is crucial to understand the impact of communication and design “targeted persuasion.” To predict the effectiveness of health advisories we need to know who will follow them.

  2. What information should governments collect and how should this be disseminated? Governments face what is called a problem of "information design", studied by an extensive recent literature (see, for example, Kamenica and Gentzkow (2011), Bergemann, and Morris (2016) and for a recent survey Bergemann, and Morris (2019)). This asks how an agent (e.g., government) should collect and communicate information to persuade others (e.g., citizens). Government wants to release the most convincing evidence without lying, but agents understand this and interpret the evidence accordingly. Theoretical analysis shows that governments can do better than by revealing all information (or statistics ex-post). Rather, governments should commit ex-ante to what information to reveal (regardless of whether this supports or damages the government’s case) and to “bunching together” some pieces of information. Properly designed, information of this type is optimal. To our knowledge no paper has estimated the effectiveness of such methods in practice. Our experiment fills this gap.

  3. Are some subjects more responsive to the opinion of their peers than experts? It is crucial to investigate if this happens and to identify areas of the society where “clusters” of inaccurate opinions form and persist.

We focus on opinions on two key economic variables: the Dow Jones index and the unemployment rate. The former matters for avoiding a self-fulfilling stock market crash; the latter is a key measure of economic health, which matters for many business decisions.

As we focus on optimal information collection by the governments to persuade citizens, with the goal of avoiding coordination failures, we believe that our work fits the Keynes Fund’s objectives very well. We seek to better understand how market failures can be avoided, with the goal of improving public policy. Moreover, our questions pertain to important questions about financial markets, and how the “animal spirits” that govern them can be reined in.



Professor Matthew Elliott


Matthew Elliott is Professor of Economics at the Faculty of Economics, University of Cambridge. His research expertise is in Networked Markets, Network Games, Bargaining, Cascades on Networks, Search, Matching.


Professor Pietro Ortoleva


Pietro Ortoleva is Professor of Economics and Public Affairs at the Department of Economics and School of Public and International Affairs, Princeton University. His research interests are Decision Theory, Behavioral Economics, Experimental Economics, Political Economy


Professor Marina Agranov


Marina Agranov is Professor of Economics at the Division of Humanities and Social Sciences, California Institute of Technology. Her research interests are Experimental Economics, Political Economics, Microeconomics.


Share this Project